Customer service coordination management system and method based on data matching

Abstract

The application discloses a customer service cooperation management system and method based on data matching, and belongs to the technical field of customer service cooperation management. The application receives customer request data from customer service channels, carries out normalization arrangement and cross-channel semantic alignment processing, extracts candidate work order data based on standard work order characteristic data and the time sequence connection existing with the current customer request, performs cross-node flow transfer tracking processing on the current work order, synchronously updates the work order state, and obtains continuous work order tracking results corresponding to the current customer request, which is beneficial to solving the problems that in the enterprise customer-oriented after-sales service scene where similar requests are repeatedly submitted by customers, multiple processing departments are involved in parallel, or the same matter is continuously tracked across time periods, the work order association recognition capability is usually insufficient, the processing stage tracking continuity is insufficient, data is split, the state is distorted, and tracking data is synchronously delayed in the work order flow transfer process, and the technical requirements of the customer request work order fine-grained state tracking cannot be met.

Description

Technical Field

[0001] This invention belongs to the field of customer service collaborative management technology, specifically a customer service collaborative management system and method based on data matching. Background Technology

[0002] As the number of customers and service channels continues to grow, customer service has evolved from a traditional single-window response to a collaborative management model encompassing multiple stages, including customer consultation, business acceptance, issue transfer, progress tracking, and result feedback. In this process, customer requests often involve various types of data, such as customer information, business information, historical interaction information, service resource information, and processing node information. Relying solely on manual judgment or static rules for data flow and allocation can easily lead to problems such as delayed responses, inaccurate transfers, disjointed collaboration, and duplicate processing. Furthermore, it makes it difficult to guarantee the continuity, accuracy, and traceability of the customer service process. Therefore, building an efficient collaborative management mechanism around customer service scenarios, especially how to effectively correlate customer needs, service resources, and processing paths using data matching methods, has become an important direction for improving customer service quality, optimizing service collaboration efficiency, and enhancing the company's refined service capabilities.

[0003] The customer service management process typically follows these steps: First, the customer service platform receives customer service requests from channels such as telephone, web pages, mobile devices, email, or social media, and collects corresponding basic customer information, business processing information, historical service records, problem descriptions, and current service status information. Second, the collected information undergoes standardization, field mapping, and tag extraction to form a customer service dataset that can be used for unified identification. Then, based on preset data matching rules, customer attribute association rules, business type correspondence rules, and service resource allocation rules, the customer service dataset is matched and analyzed with business processing node data, knowledge base data, and historical work order data. Business processing node data includes, but is not limited to, acceptance nodes, review nodes, dispatch nodes, escalation nodes, and return nodes. The system includes visit and completion nodes, and knowledge base data such as business rules, handling plans, and operation guidelines. Historical work order data includes, but is not limited to, historical problem types, historical transfer records, and historical processing times. This data is used to determine the service recipient, handling department, collaborating positions, and handling path corresponding to the current customer request. Based on this, the matching results are distributed to the corresponding customer service agents, business support personnel, or backend processing modules. During the service processing, the work order flow status, processing progress, feedback results, and customer response information are updated synchronously to achieve collaborative processing between multiple positions and multiple stages. Finally, the customer service results are recorded, archived, and evaluated based on the service completion status, and the processing results are written back to the customer service database for subsequent customer identification, service optimization, and matching rule adjustments.

[0004] In the aforementioned customer service management process, customer service requests are usually not completed directly with a single processing action. Instead, they rely on work orders as information carriers and process flow carriers to uniformly record and continuously track the acceptance information, assignment information, processing information, feedback information, and status information corresponding to customer requests.

[0005] For example, in after-sales service scenarios for enterprise customers, when a customer submits a service request regarding service activation changes or account permission adjustments, the platform typically generates a corresponding customer request ticket and assigns it to different positions such as customer service, technical support, operations and maintenance, or business review for collaborative processing. In this process, since customer requests often exist in various forms such as text descriptions, voice transcription results, form fields, and historical interaction records, the processing priorities, status update methods, and feedback standards of different positions also differ. This can easily lead to inaccurate information association during the generation, assignment, processing, escalation, and follow-up of the same customer request.

[0006] Especially when customers repeatedly submit similar requests, multiple processing departments intervene in parallel, or the same matter is continuously tracked across time periods, existing methods often use keyword matching-based work order classification and work order flow based on independent status field records. These methods cannot establish a temporal relationship graph between work orders, resulting in the failure to identify the semantic relationship between historical work orders and current requests. It is also difficult to continuously track the actual processing stage of work orders, which can easily lead to duplicate work order assignments, tracking omissions, and the inability to automatically identify work order ownership relationships based on data features. Consequently, the ability to maintain the consistency of work order status is insufficient.

[0007] For example, taking work order A as an example, a corporate customer submitted an account permission adjustment application via the web interface at 09:15 on August 10th, requesting that the permissions of employee account X be changed from read-only to read-write permissions. The customer also provided the customer ID, account name, business system, reason for application, and contact person information. Upon receiving this request, the customer service platform generates work order A, recording in it the customer ID C001, the type of matter, the account permission adjustment, the business object account X, the permission scope changed from read-only to read-write, the submission time August 10th 09:15, and the current status "Pending Processing." Subsequently, customer service personnel process work order A and forward it to the business review node, updating its status to "Under Review and Processing."

[0008] Before work order A was completed, the enterprise customer contacted customer service by phone at 10:02 AM on August 10th, requesting additional permissions for account X, including read / write access, export access, and approval access. The phone call was transcribed into new request data. The customer service platform generated work order B based on this request data, recording the customer ID C001, the request description (application to adjust account X permissions and add export and approval permissions), the submission time (August 10th, 10:02 AM), and the current status (pending acceptance). Work order A primarily sourced from web form fields, which were relatively standardized. Work order B, however, mainly derived from the phone call transcription, using different descriptions such as permission activation, permission supplementation, and additional authorization. Furthermore, work order A recorded the initial application status, while work order B recorded a supplementary submission status. The two work orders differ in field representation, content granularity, and status update time.

[0009] If existing methods only use keyword matching to categorize work orders—for example, judging work order categories solely based on terms like "permission adjustment," "account X," and "application for change"—and only rely on the independently recorded acceptance, review, and completion statuses of work orders A and B for subsequent processing, the system can typically only identify that both work orders A and B belong to the account permission category. However, it struggles to further identify that work order B is actually supplementary request data for the corresponding matter in work order A. Consequently, work order A remains at the business review stage, while work order B may be reassigned to the customer service processing stage or even re-enter the review queue. Summary of the Invention

[0010] To address the above problems, this invention provides a customer service collaborative management system and method based on data matching. The technical solution adopted by this invention is as follows:

[0011] On the one hand, a customer service collaborative management system based on data matching is provided, including: a work order feature generation and management module, a customer request time sequence connection module, and a customer request tracking and management module. Among them, the work order feature generation and management module is used to receive customer request data from customer service channels, perform normalization and arrangement and cross-channel semantic alignment processing to generate standard work order feature data; the customer request time sequence connection module is used to extract candidate work order data that have a time sequence connection with the current customer request based on the standard work order feature data; the customer request tracking and management module is used to perform cross-node flow tracking processing on the current work order based on the candidate work order data, and synchronously update the work order status to obtain continuous work order tracking results corresponding to the current customer request.

[0012] On the other hand, a data-matching-based collaborative management method for customer service is provided, including: S1, receiving customer request data from customer service channels, performing normalization orchestration and cross-channel semantic alignment processing to generate standard work order feature data; S2, based on the standard work order feature data, extracting candidate work order data that has a temporal connection with the current customer request; S3, based on the candidate work order data, performing cross-node flow tracking processing on the current work order, synchronously updating the work order status to obtain continuous work order tracking results corresponding to the current customer request.

[0013] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0014] 1. By receiving customer request data from customer service channels, normalizing and arranging it, and performing cross-channel semantic alignment processing, standard work order feature data is generated. This helps reduce data fragmentation caused by differences in fields and inconsistent semantic expressions between different channels. Based on the standard work order feature data, candidate work order data that has a temporal connection with the current customer request is extracted, which helps improve the accuracy of correlation identification between historical work orders and the current request. Based on the candidate work order data, cross-node flow tracking processing is performed on the current work order, and the work order status is updated synchronously to obtain continuous work order tracking results corresponding to the current customer request. This helps to strengthen the status continuity of work orders in multi-position collaborative processing, thereby improving the traceability and processing efficiency of the entire work order process.

[0015] 2. By performing unified field mapping processing on customer request data from different customer service channels, including channel credibility marking and field completeness marking, the system obtains mapped work order feature data. It also performs time-series connection analysis on mapped work order feature data within a preset time interval. Compared with the shortcomings of existing methods that are prone to introducing noise and incomplete data, this method helps to ensure the quality of work order features from the dual dimensions of source credibility and content completeness, and achieves accurate aggregation and time-series association of similar requests across channels.

[0016] 3. Based on standard work order feature data, candidate work order data that are sequentially connected to the current customer request are extracted, and a set of sequential connection data is further obtained. When the values ​​corresponding to the set of sequential connection data are all within the corresponding qualified sequential connection range, cross-node flow tracking processing is performed on the current work order. Compared with the shortcomings of existing methods that cannot build sequential relationships and are prone to sequential disorder and false association, this method is conducive to realizing the quantitative judgment and reliable association of sequential relationships between work orders, and improving the authenticity and continuity of work order flow tracking.

[0017] 4. When enterprise customers submit multiple requests for the same account permission adjustment within a short period of time via web, telephone, and email, and a server switch occurs simultaneously, the system analyzes the status progression relationship between adjacent processing nodes of the current work order to obtain corresponding analysis results. Based on the analysis results, it decides whether to perform node-by-node tracking processing on the current work order. Compared with the shortcomings of existing methods, such as status distortion and tracking interruption due to server switch, its advantage lies in its ability to quickly identify multiple requests for the same matter and maintain status continuity, automatically correct abnormal node jumps, and ensure that work orders can still be stably tracked, with no status loss and uninterrupted flow even in complex scenarios. Attached Figure Description

[0018] Figure 1 A schematic diagram of the structure of a data-matching-based customer service collaborative management system provided in an embodiment of this invention application;

[0019] Figure 2 This is a schematic diagram of channel credibility marking provided in an embodiment of the present invention.

[0020] Figure 3 A flowchart illustrating a data-matching-based customer service collaborative management method provided in an embodiment of this invention.

[0021] Figure 4 The divergence function curve provided in the embodiment of this invention application. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solution, the present invention will be described in detail below with reference to embodiments. The description in this part is only exemplary and explanatory, and should not be used to limit the scope of protection of the present invention in any way.

[0023] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the present invention. These examples are merely for the purpose of helping to understand the method and core ideas of the present invention. The above are only preferred embodiments of the present invention. It should be pointed out that, due to the limitations of written expression and the objective existence of infinite specific structures, those skilled in the art can make several improvements, modifications, or variations without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, variations, or combinations, or the direct application of the concept and technical solution of the present invention to other situations without modification, should all be considered within the scope of protection of the present invention.

[0024] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0025] Example 1, such as Figure 1 The diagram shown is a structural schematic of a customer service collaborative management system based on data matching provided in this application embodiment, including: a work order feature generation management module, a customer request timing connection module, and a customer request tracking management module.

[0026] The work order feature generation and management module is used in the customer service collaborative management process to receive customer request data from customer service channels, perform normalization and cross-channel semantic alignment processing, and generate standard work order feature data to represent the time evolution and status expression dimensions of the current customer request. Customer service channels include at least one of web, telephone, mobile terminal, email, or social media. For example, when the same enterprise customer submits a permission change application through the web and then supplements the permission scope description through the telephone regarding the permission adjustment of an employee account in the target business, by uniformly mapping the submission time field and processing status field in the web and the call registration time field and current acceptance status field in the telephone, it helps to generate standard work order feature data that can uniformly represent the submission sequence and status expression relationship of the same customer request under different channels, thereby providing a unified data foundation for subsequent work order association identification.

[0027] The customer request timing module is used to extract candidate work order data that are sequentially connected to the current customer request based on standard work order feature data. For example, if an enterprise customer submits an initial permission application on the web page and then submits a supplementary request via telephone within a short period of time, and the corresponding work order on the web page has already been transferred to the business review node, by extracting candidate work order data that are sequentially connected to the current customer request and comparing their time sequence, it is helpful to extract candidate work order data that are sequentially related to the current customer request, and thus determine whether the supplementary request on the telephone is a continuation of the processing of the matter corresponding to the initial work order on the web page.

[0028] The customer request tracking and management module is used to perform cross-node flow tracking processing on the current work order based on candidate work order data, and synchronously update the work order status to obtain continuous work order tracking results corresponding to the current customer request. For example, when a supplementary request from the telephone end is identified as the same permission adjustment item corresponding to the initial work order on the web end, and the corresponding candidate work order is in the review and processing node, by tracking and analyzing the status progress relationship between the current work order and the candidate work order at the customer service acceptance node, business review node, and permission execution node, it is beneficial to improve the ability to maintain the status continuity of the same customer request in the multi-node flow process.

[0029] In this embodiment, the work order feature generation and management module, the customer request timing connection module, and the customer request tracking and management module provided in this invention application are interconnected and their functions support each other. Specifically, the work order feature generation and management module, as a basic support module, generates unified standard work order feature data by normalizing and arranging multi-channel customer request data and performing cross-channel semantic alignment. This data is not only the core input for the customer request timing connection module to carry out its work, but also provides a standardized and comparable data foundation for it to realize timing connection identification. That is to say, the customer request timing connection module can match the current customer request with historical work order data based on the standard work order feature data to identify candidate work order data that correspond to the current customer request in terms of customer identifier, matter type, business object, and time position. For example, for the initial web application and the supplementary telephone application for the same permission adjustment matter of an enterprise customer, it is based on the unified feature data generated by this module that the customer request timing connection module can accurately extract the candidate work order data of the initial web application and the status data of the business review node therein, thereby completing the timing association judgment of the previous and subsequent requests.

[0030] The candidate work order data extracted and matched by the customer request timing module provides crucial information for the cross-node flow tracking of the customer request tracking management module. This allows the module to clearly define the relationship between the current work order (e.g., a supplementary request work order from a telephone terminal) and candidate work orders (e.g., an initial work order from a web terminal), as well as the current node position. This enables continuous tracking and status synchronization of the same permission adjustment across multiple nodes, including customer service acceptance, business review, and permission execution. Simultaneously, the work order status update results and node flow anomaly information generated by the customer request tracking management module during the tracking process are fed back to the work order feature generation management module. This feedback optimizes its normalization mapping rules and semantic alignment strategies, further improving the accuracy and relevance of standard work order feature data.

[0031] Preferably, customer request data from customer service channels is received and normalized and cross-channel semantic alignment is performed. The specific process is as follows: Customer request data from different customer service channels is subjected to unified field mapping processing, including channel credibility markers and field completeness markers, to obtain mapped work order feature data; combined with the time field, the mapped work order feature data within a preset time range is subjected to time sequence connection analysis to obtain standard work order feature data that can reflect the order of requests and the evolution trajectory of content; the time field represents a data field used to characterize one or more time information such as the submission time, supplement time, transfer time, status update time, or feedback time of the customer request corresponding to the matter in different customer service channels; customer request data represents data used to characterize the customer's service request to the customer service platform around the target service matter, including customer identifiers, etc.

[0032] The process of performing time-series connection analysis on mapped work order feature data within a preset time proximity interval is as follows: Mapped work order feature data whose time fields fall within the preset time proximity interval are marked as candidate time-series associated data; based on the chronological order corresponding to the time fields, the candidate time-series associated data are sorted to generate standard work order feature data that reflects the chronological order of customer request processing; the candidate time-series associated data are sorted according to one of the following chronological relationships: request submission time, status update time, supplementary feedback time, or node transition time, and the sorting result is marked as standard work order feature data.

[0033] Specifically, by performing time-series connection analysis to obtain candidate time-series related data, and sorting the candidate time-series related data based on the chronological order of the time fields, standard work order feature data is obtained. This helps to support the accurate aggregation and logical sorting of related work orders submitted through multiple channels and at multiple time periods under the same customer identifier, clarify the sequential relationship and continuity of each work order, and thus enhance the coherence and accuracy of work order time-series association.

[0034] It should be added that a unified field mapping process, including channel credibility markers and field completeness markers, is performed on customer request data from different customer service channels. The specific process is as follows:

[0035] First, the time and status fields corresponding to customer request data from different customer service channels are extracted. Based on preset field correspondence rules and semantic merging mapping algorithms, initial mapped work order feature data is obtained. For example, in one embodiment, the submission time field in the web page, the call registration time field in the telephone terminal, and the email receipt time field in the email terminal can be mapped to a unified time field. The processing status field in the web page, the current acceptance status field in the telephone terminal, and the processing progress description field in the email terminal can be mapped to a unified status field. The semantic merging mapping algorithm is used to merge semantically similar expressions such as "under review", "under approval", and "transferred for review" into a unified status semantic to obtain initial mapped work order feature data. The preset field correspondence rules represent a set of rules pre-set based on the processing flow corresponding to customer service items, used to establish the correspondence mapping relationship between heterogeneous fields and unified work order fields under different customer service channels. The status field represents an information field that characterizes one or more processing states of a customer request in the corresponding processing flow, such as acceptance status, review status, dispatch status, escalation status, follow-up status, completion status, or pending supplementation status.

[0036] Then, based on the initial mapping work order feature data, the channel credibility is statistically analyzed to obtain data that quantifies channel credibility. Initial mapping work order feature data that meets the channel credibility conditions are marked with channel credibility tags, while initial mapping work order feature data that does not meet the channel credibility conditions undergoes channel credibility enhancement processing. After processing, channel credibility tags are added. The channel credibility conditions indicate that the data that quantifies channel credibility is within the corresponding pre-set preset channel credibility range. The data that quantifies channel credibility is represented by the difference between the submission response time of the customer request corresponding to the work order in the customer service channel within the preset statistical period and the preset submission response time. The submission response time represents the duration monitored by the timer from when the customer initiates a request through the corresponding customer service channel until the preset customer service center successfully receives and generates the corresponding initial request record.

[0037] like Figure 2 The diagram illustrates the channel credibility marking provided in an embodiment of this invention: Data quantifying channel credibility is obtained; initial mapping work order feature data that meets the channel credibility conditions is marked with channel credibility; initial mapping work order feature data that does not meet the channel credibility conditions undergoes channel credibility enhancement processing, i.e., the data quantifying channel credibility is input into a preset trusted channel classification model; initial mapping work order feature data corresponding to low-credibility channels are used as low-credibility channel data; low-credibility channel data is deleted; and channel credibility marking is added to the remaining initial mapping work order feature data.

[0038] The embodiments of this invention pre-construct a database containing various ranges or preset values, such as preset channel credibility range, preset field completeness range, qualified time sequence connection range, preset submission response duration, etc.

[0039] The following example illustrates the construction process of such a range or interval: Historical quantitative channel credibility data for multiple customer service channels within a historical statistical period is collected in advance. Based on the actual situation of each customer service channel during historical work order processing, the historical quantitative channel credibility data is labeled with a credibility level to obtain a credibility channel sample set. The minimum and maximum values ​​of the quantitative channel credibility data corresponding to the credibility channel sample set are calculated. The numerical range between the minimum and maximum values ​​is determined as the initial channel credibility interval. This initial channel credibility interval is then applied to new historical customer requests for verification. When the proportion of data falling within the initial channel credibility interval reaches a preset percentage, the initial channel credibility interval is determined as the preset channel credibility range and written into the database.

[0040] Taking the preset submission response time as an example, the process of constructing such preset values ​​is as follows: The submission response time of customer requests corresponding to work orders in each customer service channel within multiple historical statistical periods is collected in advance. The submission response time is then statistically analyzed for each customer service channel, for example, the actual submission response time for web-based customer requests, telephone-based customer requests, and email-based customer requests. The actual submission response times for each type of customer service channel are sorted by size, and the smallest values ​​in the top 5% and the largest values ​​in the bottom 5% of the sorted results are removed. The average value of the removed actual submission response times is then calculated, and this average value is used as the preset submission response time for that customer service channel.

[0041] Simultaneously, based on the initial mapped work order feature data, the field completeness is statistically analyzed to obtain data quantifying field completeness. Initial mapped work order feature data meeting the field completeness criteria are marked with field completeness tags, while those not meeting the criteria undergo field completeness enhancement processing, with field completeness tags added after processing. Initial mapped work order feature data with both channel credibility and field completeness tags are statistically analyzed and used as mapped work order feature data. Field completeness criteria indicate that the data quantifying field completeness falls within the corresponding pre-set preset field completeness range. The data quantifying field completeness is represented by the number of fields identified in the customer service channel within a preset statistical period. The number of identified fields represents the number of fields in the current work order record monitored by the counter that can be identified and mapped to the preset work order field set. The preset work order field set represents a pre-defined set of data fields based on the processing flow corresponding to customer service matters, used to characterize the customer identification information, matter type information, business object information, time information, status information, and feedback information required to be recorded during the generation, acceptance, transfer, review, feedback, and completion of the work order.

[0042] In this embodiment, by performing unified field mapping processing on customer request data from different customer service channels, mapped work order feature data is obtained. Combined with the time field, time-series connection analysis is performed on the mapped work order feature data within a preset time range. This helps improve the accuracy of time-series association identification of the same customer request in multi-channel submission scenarios, avoids work order association omissions or misjudgments caused by channel heterogeneity and inconsistent fields, and further helps to achieve accurate aggregation of work orders related to the same matter submitted by customers across channels and time periods, providing coherent and complete work order data support for subsequent cross-node flow tracking.

[0043] This method acquires quantitative channel credibility data based on initial mapped work order feature data and statistically analyzes initial mapped work order feature data with both channel credibility and field completeness tags added. Compared to existing technologies that easily introduce invalid or low-quality data, leading to lower accuracy in subsequent work order association and tracking, this method helps to achieve quality control of mapped work order feature data, eliminating invalid, low-credibility, and missing field work order data. This is more conducive to improving the accuracy of subsequent time-series connection analysis and cross-node tracking processing, ensuring the reliability and consistency of data throughout the entire work order flow process, and further optimizing the processing efficiency and service quality of customer request work orders.

[0044] Preferably, the channel credibility enhancement process involves performing a credibility ranking on the initial mapping work order feature data corresponding to the same customer request under different customer service channels, and deleting low-credibility channel data. The specific process is as follows: input the data that quantifies channel credibility into a preset credibility channel classification model, output credibility channel classification results including high-credibility channels and low-credibility channels, use the initial mapping work order feature data corresponding to low-credibility channels as low-credibility channel data, delete the low-credibility channel data, and add channel credibility tags to the remaining initial mapping work order feature data, that is, the initial mapping work order feature data corresponding to high-credibility channels.

[0045] The field completeness enhancement process involves sorting the feature data of initial mapping work orders corresponding to the same customer request under different customer service channels based on their completeness, and deleting data with low completeness. The specific process is as follows: The data quantifying field completeness is input into a preset field completeness classification model, and the output includes field completeness classification results including high completeness fields and low completeness fields. The initial mapping work order feature data corresponding to the low completeness fields is taken as low completeness data, and the low completeness data is deleted. The remaining initial mapping work order feature data, that is, the initial mapping work order feature data corresponding to the high completeness fields, are marked with field completeness.

[0046] It should be explained that the preset trusted channel classification model and the preset field completeness classification model are models trained based on existing machine learning algorithms, which have the functions of classification and result output. The training data for the preset trusted channel classification model consists of data that quantifies channel trustworthiness and pre-set trusted channel classification results, while the training data for the preset field completeness classification model consists of data that quantifies field completeness and pre-set field completeness classification results.

[0047] Taking a specific machine learning algorithm and a pre-defined trusted channel classification model as an example, the construction process is illustrated as follows: An existing support vector machine (SVM) algorithm is selected as the training algorithm. Data on the quantified channel trustworthiness of multiple customer service channels within a pre-defined statistical period is collected and used as training feature data. Simultaneously, based on historical processing results, channel stability, and state consistency, trusted channel classification results are pre-defined for each training sample. Then, the training samples labeled with trusted channel classification results are divided into a training set and a validation set. The training set is input into the SVM training framework. The data on quantified channel trustworthiness is mapped to the corresponding classification feature space using a kernel function mapping method. The position and margin parameters of the classification hyperplane are iteratively adjusted to gradually increase the classification margin between high-trust channel samples and low-trust channel samples until a pre-defined convergence condition is met, thus obtaining the initial trusted channel classification model.

[0048] Next, the data on the quantified channel credibility in the validation set is input into the initial credible channel classification model. Based on the consistency between the output credible channel classification results and the preset labeling results, the kernel function parameters, penalty parameters, and classification boundary parameters of the initial credible channel classification model are tuned until the model output meets the preset classification accuracy requirements, thus obtaining the preset credible channel classification model. Finally, after the model is built, the data on the quantified channel credibility is input into the preset credible channel classification model, and the corresponding credible channel classification results are output to distinguish between high-credibility channels and low-credibility channels.

[0049] In this embodiment, by performing channel credibility enhancement processing and deleting low-credibility channel data when the channel credibility condition is not met, and by performing field integrity enhancement processing and deleting low-completeness data when the field integrity condition is not met, compared with the shortcomings of existing technologies such as large noise in work order features, easy error in association identification, and distorted tracking results, it helps to improve the overall reliability and standardization of mapped work order feature data. At the same time, it can effectively filter invalid, distorted, and incomplete data, reduce the misjudgment rate of subsequent time-series association analysis and cross-node tracking, and ensure the accuracy and continuity of work order status tracking.

[0050] It should be further explained that the channel credibility enhancement process and the field completeness enhancement process are not independent of each other. Specifically, channel credibility directly affects the effectiveness and accuracy of field collection. Low credibility channels are often accompanied by problems such as missing fields and field errors, while insufficient field completeness will further reduce the credibility of work order data. Improving both simultaneously can jointly optimize the quality of work order features from two dimensions: data source reliability and content completeness, and achieve mutual verification and mutual reinforcement.

[0051] Preferably, based on standard work order feature data, candidate work order data that has a temporal connection with the current customer request is extracted. The specific process is as follows: the submission time, node flow time, and feedback time corresponding to the standard work order feature data are compared to determine the temporal sequence, and a set of temporal connection data representing the qualified status of the customer request temporal connection is extracted. The set of temporal connection data is used as the benchmark judgment data for whether to trigger cross-node flow tracking processing for the current work order: when the values ​​corresponding to the set of temporal connection data are all within the pre-set qualified temporal connection range, the corresponding candidate work order data is extracted, and cross-node flow tracking processing for the current work order is triggered; otherwise, cross-node flow tracking processing for the current work order is not triggered, and a temporal connection alarm is sent to the preset data management center. The candidate work order data includes historical work order number, historical work order customer identifier, historical work order submission time, etc.

[0052] When the data values ​​in the time-series connection data set are not within the corresponding range, it indicates that there is a time-series mismatch between the current customer request and the candidate work order in at least one of the following: submission time connection, processing node advancement, or feedback response. This means that a continuous processing link that meets the preset conditions has not been formed between the current customer request and the candidate work order, or there are time-series anomalies such as abnormal delays, stage jumps, state inversions, and abnormal interruptions. Therefore, it is not appropriate to directly determine the current customer request as the continuation processing data of the corresponding item of the candidate work order. The time-series connection data set includes submission time interval data that represents the interval relationship between the submission time of the current customer request and the submission time of the candidate work order, node flow time interval data that represents the stage succession relationship between the node flow time of the current customer request and the node flow time of the candidate work order, and feedback time interval data that represents the feedback response status of the current customer request. The candidate work order refers to the historical work order selected from the historical work order data based on the customer identifier, item type, and business object corresponding to the current customer request.

[0053] It should be added that the specific process for obtaining the time-series data set is as follows: the time difference between the submission duration of the current customer request monitored by the timer and the submission duration of the candidate work order is used as the submission time interval data; the time difference between the node flow duration of the current customer request monitored by the timer and the status update duration of the candidate work order is used as the node flow time interval data; and the time difference between the feedback duration of the current customer request monitored by the timer and the completion feedback duration of the candidate work order is used as the feedback time interval data.

[0054] It should be explained that the data in the time-series connection data set do not exist independently, but are used together to characterize the overall time-series connection relationship between the current customer request and the candidate work order in the submission stage, node flow stage, and feedback stage. The data complement and corroborate each other, and can comprehensively characterize whether there is a continuous processing relationship between the current customer request and the candidate work order from the request initiation, processing to result feedback, rather than making isolated judgments based on a single time difference. This helps to improve the accuracy and stability of the time-series connection analysis results.

[0055] In this embodiment, by extracting candidate work order data that has a temporal connection with the current customer request—that is, when the values ​​corresponding to the temporal connection data set are all within the corresponding qualified temporal connection range—the corresponding candidate work order data is extracted; otherwise, a temporal connection alarm is sent to the preset data management center. This helps to accurately determine the temporal relationship between the current customer request and historical work orders, reduce the incorrect association of irrelevant work orders, and promptly detect and warn of work order matching behaviors with abnormal timing, excessive span, or no connection, thereby reducing the probability of association distortion or tracking misalignment in subsequent workflow tracking.

[0056] Preferably, based on candidate work order data, cross-node flow tracking processing is performed on the current work order, and the work order status is updated synchronously to obtain continuous work order tracking results corresponding to the current customer request. Specifically, this involves analyzing the status progression relationship of the current work order between adjacent processing nodes, obtaining corresponding analysis results, and determining whether to perform node-by-node tracking processing on the current work order based on the analysis results. The specific process is as follows: the node dwell time, i.e., the time the work order takes from entry to exit at a preset processing node, is used as the node flow analysis result; when the node flow analysis result is within the corresponding qualified flow analysis range, indicating that there is a continuous progression relationship between the current work order and multiple processing nodes, node-by-node tracking processing is performed on the current work order to generate the corresponding node tracking results; when the node flow analysis result is not within the corresponding qualified flow analysis range, indicating that the current work order has duplicate entry nodes, abnormal node jumps, state inversion, or missing feedback. In the event of at least one of the following situations: interruption of workflow, the current corresponding customer service node type is determined to be an abnormal customer service node, and the abnormal customer service node is replaced. After the replacement is completed, the node workflow analysis results are re-monitored. If the node workflow analysis results are within the corresponding pre-set qualified workflow analysis range, the current work order is processed by tracking each node; otherwise, a node workflow abnormality alarm is triggered. Replacing an abnormal customer service node means selecting the customer service node whose workflow analysis results are within the corresponding qualified workflow analysis range within a preset service time period, using the selected customer service node as the target replacement node, and replacing the current abnormal customer service node based on the target replacement node. A customer service node refers to the business processing node corresponding to the work order in the customer service processing process, including one or more of the following: acceptance node, review node, dispatch node, escalation node, follow-up node, and completion node.

[0057] Specifically, this involves performing node-by-node tracking on the current work order and generating corresponding node tracking results. This means: following a pre-defined work order flow path, data is collected from each customer service node, recording the work order information for each node and uploading it as a node tracking result to a pre-defined data management center. The work order information includes the node name corresponding to the current customer service node, the time of entry into the node, and the time of exit from the node. The pre-defined work order flow path represents a pre-set node path based on the processing flow corresponding to the customer service item, used to characterize the flow order and succession relationship of the work order among multiple customer service nodes. Performing node-by-node tracking on the current work order generates… The corresponding node tracking results also include synchronized updates to the work order status. Synchronized updates to the work order status mean that the work order processing stage, node affiliation status, and processing progress information are refreshed in real time to maintain the consistency and continuity of status data during the cross-node flow of the work order. The work order processing stage represents the business stage of the current work order in the processing links such as acceptance, review, dispatch, escalation, follow-up, and completion. The node affiliation status represents the information of the customer service node corresponding to the current work order. The processing progress information represents the information of the completed nodes and pending nodes in the customer service process from the time the current work order was generated to the current time.

[0058] like Figure 3 The flowchart shown is a data matching-based customer service collaborative management method provided in an embodiment of the present invention, including: S1, receiving customer request data from customer service channels during the customer service collaborative management process, performing normalization and cross-channel semantic alignment processing to generate standard work order feature data to characterize the time evolution dimension and status expression dimension of the current customer request. The customer service channels include at least one of web, telephone, mobile terminal, email, or social media. S2, extracting candidate work order data that has a temporal connection with the current customer request based on the standard work order feature data. S3, performing cross-node flow tracking processing on the current work order based on the candidate work order data, synchronously updating the work order status to obtain continuous work order tracking results corresponding to the current customer request.

[0059] In this embodiment, the node flow analysis results are obtained by analyzing the state progression relationship between adjacent processing nodes of the current work order. When the node flow analysis results are within the corresponding qualified flow analysis range, node-by-node tracking processing is performed on the current work order. Otherwise, the abnormal customer service node is replaced based on the target replacement node. This helps to improve the continuity and state consistency of the work order flow across nodes, while strengthening the automatic correction capability for problems such as flow abnormalities, node blocking, and state inversion. It also reduces work order tracking interruptions or state distortions and helps to ensure the traceability of the entire life cycle of the work order.

[0060] Example 2, based on the method provided in Example 1, involves a scenario where an enterprise customer submits multiple requests for adjusting the same account permissions within a short period via web, telephone, and email, while a server switch occurs simultaneously. The method analyzes the status progression of the current work order across adjacent processing nodes to obtain corresponding analysis results. Based on these results, a decision is made on whether to perform node-by-node tracking processing on the current work order. This further includes: when the node flow analysis result is within the corresponding qualified flow analysis range, performing node-by-node tracking processing on the current work order and generating corresponding node tracking results; when the node flow analysis result is outside the corresponding qualified flow analysis range, classifying the current customer service node type as an abnormal customer service node, performing node replacement based on the central processing unit and historical qualified nodes, and writing the replacement node's status into the target work order status area in memory. This information is then sent to the corresponding customer service terminal and the review department via a message bus. The core terminal is updated synchronously, and the node flow analysis results are re-monitored. If the node flow analysis results are within the corresponding qualified flow analysis range, the current work order is processed by tracking each node; otherwise, a node flow abnormality alarm is triggered. Historical qualified nodes refer to the nodes corresponding to each customer service node that are pre-recorded and verified to meet the qualified flow analysis range during the normal flow of historical work orders. The target work order status area refers to the memory area in the computer memory that is pre-allocated to temporarily store the current work order after node replacement. Customer service terminals refer to terminal devices used to receive customer service requests, enter customer interaction information, display work order processing status, and perform customer service acceptance operations, including telephone acceptance terminals, self-service terminals, and customer service agent terminals. Review terminals refer to terminal devices used to perform business review, permission approval, processing verification, or result confirmation on the corresponding matters of the work order, including permission approval terminals.

[0061] In this embodiment, when an enterprise customer submits multiple requests for adjusting the same account permissions via web, telephone, and email within a short period, and a server switch occurs simultaneously, the node flow analysis results are obtained by analyzing the status progression relationship between adjacent processing nodes of the current work order. If the node flow analysis results are within the corresponding qualified flow analysis range, node-by-node tracking processing is performed on the current work order; otherwise, node replacement is performed based on the central processor and historical qualified nodes. This helps improve the accuracy of abnormal node identification and the ability to reconstruct the work order flow chain, thereby ensuring the continuity of the current work order's status during cross-node processing and reducing the risk of status misalignment, abnormal node jumps, and duplicate order dispatch caused by server switch. Furthermore, synchronous updates to the corresponding customer service terminal and review terminal via the message bus help improve the efficiency of work order status synchronization and the consistency of multi-terminal status. At the same time, it helps shorten the information transmission latency between the customer service terminal and the review terminal, ensuring that each processing node has the ability to perceive and coordinate the progress of the current work order in real time.

[0062] like Figure 4 The figure shows a divergence function curve provided in an embodiment of the present invention: obtaining the matching deviation between the customer request feature value and the reference state. The customer request feature value refers to the numerical feature formed after encoding and normalizing customer request information, such as work order information. The horizontal axis represents the customer request feature value or matching index, and the vertical axis represents the matching deviation value. Different colored curves in the figure correspond to different beta values. As the customer request feature value deviates from the reference state, the divergence value increases.

Claims

1. A customer service coordination management system based on data matching, characterized by, include: The module includes a work order feature generation and management module, a customer request timing connection module, and a customer request tracking and management module. The work order feature generation and management module is used to receive customer request data from customer service channels, perform normalization and cross-channel semantic alignment processing, and generate standard work order feature data. The customer request timing connection module is used to extract candidate work order data that has a timing connection with the current customer request based on the standard work order feature data; The customer request tracking management module is used to perform cross-node flow tracking processing on the current work order based on the candidate work order data, and synchronously update the work order status to obtain the continuous work order tracking results corresponding to the current customer request.

2. The customer service collaboration management system based on data matching according to claim 1, characterized in that, The process of receiving customer request data from customer service channels, performing normalization and cross-channel semantic alignment, is as follows: The customer request data from different customer service channels is subjected to unified field mapping processing, which includes channel credibility markers and field completeness markers, to obtain mapped work order feature data; By combining the time field, a time-series connection analysis is performed on the mapped work order feature data within a preset time interval to obtain standard work order feature data that can reflect the order of requests and the evolution trajectory of content.

3. The customer service collaboration management system based on data matching according to claim 2, characterized in that, The specific process of performing unified field mapping processing on customer request data from different customer service channels, including channel credibility markers and field completeness markers, is as follows: Extract the time and status fields corresponding to customer request data from different customer service channels, and map them based on preset field correspondence rules and semantic merging mapping algorithm to obtain initial mapped work order feature data; Based on the initial mapping work order feature data, the channel credibility is statistically analyzed, and data on the quantitative channel credibility is obtained. The initial mapping work order feature data that meets the channel credibility conditions is marked with a channel credibility tag. The initial mapping work order feature data that does not meet the channel credibility conditions is processed to improve channel credibility. After the processing is completed, a channel credibility tag is added. The data for quantifying channel credibility is represented by the difference between the submission response time of the corresponding work order for customer requests in the customer service channel within a preset statistical period and the preset submission response time; Based on the initial mapping work order feature data, the field completeness is statistically analyzed to obtain data on the quantitative field completeness. Field completeness tags are added to the initial mapping work order feature data that meet the field completeness conditions, and field completeness improvement processing is performed on the initial mapping work order feature data that do not meet the field completeness conditions. Field completeness tags are added after the processing is completed. The initial mapping work order feature data with both channel credibility and field completeness tags added simultaneously is used as the mapping work order feature data. The data on the completeness of the quantified fields is represented by the number of fields identified in the customer service channels within a preset statistical period.

4. The customer service collaboration management system based on data matching according to claim 3, characterized in that, The channel credibility enhancement process involves performing a credibility ranking on the initial mapping work order feature data corresponding to the same customer request under different customer service channels, and deleting low-credibility channel data. The specific process is as follows: Input the data on the credibility of the quantified channel into the preset credible channel classification model, output the credible channel classification result, use the initial mapped work order feature data corresponding to the low-credible channel as the low-credible channel data, and delete the low-credible channel data. The field completeness improvement process involves sorting the feature data of the initial mapping work orders corresponding to the same customer request under different customer service channels based on their completeness, and deleting data with low completeness. The specific process is as follows: The data on the quantified field completeness is input into the preset field completeness classification model, and the field completeness classification result is output. The initial mapped work order feature data corresponding to the low completeness field is used as the low completeness data, and the low completeness data is deleted.

5. The customer service collaboration management system based on data matching according to claim 2, wherein, The specific process for performing time-series connection analysis on the mapped work order feature data within a preset time interval is as follows: Mapped work order feature data whose time field falls within a preset time proximity interval are marked as candidate time-series related data; Based on the chronological order of the time fields, the candidate time-series related data are sorted to generate standard work order feature data; The candidate time-series associated data is sorted according to one of the following chronological relationships: request submission time, status update time, supplementary feedback time, or node transition time. The sorting result is then marked as standard work order feature data.

6. The customer service collaboration management system based on data matching according to claim 1, wherein, The process of extracting candidate work order data that has a temporal connection with the current customer request based on the standard work order feature data is as follows: Based on the submission time, node flow time and feedback time corresponding to the standard work order feature data, the temporal sequence relationship is compared, and a time-series data set is extracted. The time-series data set is used as the benchmark for determining whether to trigger cross-node flow tracking processing for the current work order: When the values ​​corresponding to the time sequence connection data set are all within the corresponding qualified time sequence connection range, the corresponding candidate work order data is extracted, and cross-node flow tracking processing is triggered for the current work order; otherwise, cross-node flow tracking processing is not triggered for the current work order, and a time sequence connection alarm is sent to the preset data management center. The time-series connection data set includes submission time interval data, node flow time interval data, and feedback time interval data; The specific process for obtaining the time-series data set is as follows: The time difference between the submission duration of the current customer request and the submission duration of the candidate work order is used as the submission time interval data. The time difference between the node processing time corresponding to the current customer request and the status update time corresponding to the candidate work order is used as the node processing time interval data. The time difference between the feedback time corresponding to the current customer request and the completion feedback time corresponding to the candidate work order is used as the feedback time interval data.

7. The customer service collaboration management system based on data matching according to claim 1, characterized in that, Based on the candidate work order data, cross-node flow tracking processing is performed on the current work order, and the work order status is updated synchronously to obtain continuous work order tracking results corresponding to the current customer request. Specifically, this involves analyzing the status progression relationship of the current work order between adjacent processing nodes, obtaining corresponding analysis results, and determining whether to perform node-by-node tracking processing on the current work order based on the analysis results. The specific process is as follows: The duration of node dwell time is used as the result of node flow analysis; When the node flow analysis result is within the corresponding qualified flow analysis range, perform node-by-node tracking processing on the current work order and generate the corresponding node tracking result; When the node flow analysis result is not within the corresponding qualified flow analysis range, the current corresponding customer service node type is judged as an abnormal customer service node, and the abnormal customer service node is replaced. After the replacement is completed, the node flow analysis result is re-monitored. If the node flow analysis result is within the corresponding qualified flow analysis range, the current work order is processed by node-by-node tracking; otherwise, a node flow abnormality alarm is triggered. The replacement of abnormal customer service nodes means selecting the customer service node whose node flow analysis results are within the corresponding qualified flow analysis range within a preset service time period, using the selected customer service node as the target replacement node, and replacing the current abnormal customer service node based on the target replacement node.

8. The customer service collaboration management system based on data matching according to claim 7, characterized in that, The step of performing node-by-node tracking processing on the current work order and generating corresponding node tracking results specifically means: collecting data from each customer service node according to the preset work order flow link, recording the work order information of each node, and uploading it to the preset data management center as the node tracking result; The step of performing node-by-node tracking processing on the current work order and generating corresponding node tracking results also includes synchronously updating the work order status. The synchronous update of the work order status means refreshing the work order processing stage, node ownership status and processing progress information in real time.

9. The customer service collaboration management system based on data matching according to claim 7, characterized in that, The process of analyzing the state progression relationship of the current work order between adjacent processing nodes, obtaining corresponding analysis results, and determining whether to perform node-by-node tracking processing on the current work order based on the analysis results also includes: When the node flow analysis result is within the corresponding qualified flow analysis range, perform node-by-node tracking processing on the current work order and generate the corresponding node tracking result. When the node flow analysis result is not within the corresponding qualified flow analysis range, the current customer service node type is determined to be an abnormal customer service node. Based on the central processing unit and historical qualified nodes, node replacement is performed, and the ownership status of the replaced node is written into the target work order status area in memory. The node flow analysis result is updated synchronously to the corresponding customer service terminal and the review terminal through the message bus. The node flow analysis result is re-monitored. If the node flow analysis result is within the corresponding qualified flow analysis range, node-by-node tracking processing is performed on the current work order. Otherwise, a node flow abnormality alarm is triggered. The customer service terminal refers to a terminal device used to receive customer service requests, input customer interaction information, display work order processing status, and perform customer service acceptance operations. The review terminal refers to a terminal device used to perform business review, permission approval, processing verification, or result confirmation on the items corresponding to the work order.

10. The method for use in the data-matching-based customer service collaboration management system of any one of claims 1-9, characterized in that, include: S1 receives customer request data from customer service channels, performs normalization and cross-channel semantic alignment processing, and generates standard work order feature data. S2, Based on the standard work order feature data, extract candidate work order data that has a temporal connection with the current customer request; S3. Based on the candidate work order data, perform cross-node flow tracking processing on the current work order and update the work order status synchronously to obtain the continuous work order tracking result corresponding to the current customer request.

Images